CN110082429A - A kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning - Google Patents
A kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning Download PDFInfo
- Publication number
- CN110082429A CN110082429A CN201910317985.3A CN201910317985A CN110082429A CN 110082429 A CN110082429 A CN 110082429A CN 201910317985 A CN201910317985 A CN 201910317985A CN 110082429 A CN110082429 A CN 110082429A
- Authority
- CN
- China
- Prior art keywords
- tunnel
- machine learning
- destructive testing
- combination machine
- reflection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/04—Analysing solids
- G01N29/045—Analysing solids by imparting shocks to the workpiece and detecting the vibrations or the acoustic waves caused by the shocks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2291/00—Indexing codes associated with group G01N29/00
- G01N2291/02—Indexing codes associated with the analysed material
- G01N2291/023—Solids
- G01N2291/0232—Glass, ceramics, concrete or stone
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Chemical & Material Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Acoustics & Sound (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Biochemistry (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Analytical Chemistry (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Pathology (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Immunology (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
Abstract
The invention discloses a kind of auxiliary judgement methods of the tunnel-liner non-destructive testing of combination machine learning: extracting characteristic parameter from hitting in vibration signal for acquisition；Regression fit is carried out to the reflection interval in characteristic parameter, obtains calibration value；Original signal is indicated with obtained characteristic parameter, this group of characteristic value is marked, records its defect situation, in this, as a training set；Step above-mentioned steps are repeated in different test objects, increase training set quantity；Model training is carried out using model training software；By the model built up, the data of unknown testing result are parsed.The present invention has cut down the adverse effect generated due to thickness, material variation, the reflection interval at the lining cutting back side is increased as determining defects parameter, it can preferably reflect defect characteristic, it solves existing detection method to be affected by staff's subjective factor, the poor problem of detection accuracy, it realizes and improves detection accuracy, the interference of reduction personnel's subjectivity, it is ensured that the objective and accurate effect of testing result.
Description
Technical field
The present invention relates to civil engineering detection fields, and in particular to a kind of tunnel-liner non-destructive testing of combination machine learning it is auxiliary
Help determination method.
Background technique
China region is vast in territory, and geographical conditions are also complicated and changeable, and the construction in tunnel is mostly in high mountain jungle, Er Qieshou
The factors such as geological environment, construction environment, construction technology, design, management influence, if construction level is inadequate, or construction
Technique it is lack of standardization and or process is not stringent and shortcoming management etc., easily cause thickness of tunnel linings insufficient, lining cutting and country rock
It contacts leakiness, cavity, lining deformation rhegma, leak, even chip off-falling occur, these problems are difficult to be sent out at tunnel construction initial stage
Show, and will lead to the bearing capacity of tunnel lining structure and durability constantly to reduce, these quality problems leave to tunnel safety
Huge hidden danger, examination and improvement not in time will seriously affect engineering be delivered for use and operation security, so in order to ensure
The operation security in tunnel carries out detection and administers as early as possible to be to be highly desirable using effective means to existing defect.
In the prior art, for the detection method of tunnel-liner internal flaw have geological radar method, wave velocity method, IAE method,
Equivalent wave velocity method etc..
Ground Penetrating Radar is a kind of lossless detection method that dielectric distribution situation is definitely descended using reflection of electromagnetic wave.Pass through hair
It penetrates antenna transmitting frequency electromagnetic waves and reflection echo is generated, by receiving antenna when electromagnetic wave encounters the different interface of dielectric constant
Reflection echo is received, determines reflecting interface away from top surface by analyzing time and the form of the reflection echo received
Distance and determine reflector a possibility that matter.Wherein, determine reflection whether factor depend on material permittivity ε (or
Velocity of wave) variation.Meanwhile the frequency of electromagnetic wave also affects investigation depth and resolution ratio.In general, frequency is lower, investigation depth
It is deeper and resolution ratio is lower.
Impact elasticity wave hammer or the impact of other exciting devices generate, and can directly reflect the mechanical characteristic of material, have
Hammer vibration energy is big, it is easy to operate, when being convenient for the features such as spectrum analysis wave velocity method test defect in concrete, it is main according to bullet
The reflection, diffraction characteristic of property wave, that is, equivalent wave velocity method.Move towards direction along survey line, carried out point by point in the form of scanning exciting and
Receive signal.The presence or absence of signal, power, propagation time etc. are reflected from defective locations and bottom position by analysis accumulation signal
Characteristic, to judge the presence or absence of defect in test point lower concrete.This method detection accuracy is high, resolving power is strong, the scope of application compared with
Extensively.Its feature has: (1) elastic wave of exciting can generate reflection (theoretical basis of IE method) in fault location；(2) elastic wave of exciting
It is longer than zero defect position that the time used is reflected in defective position.Therefore, equivalent velocity of wave just seem it is slower (IEEV method
Theoretical basis).The test scope of this method is deeper, up to 1 meter or more, however, impact elasticity wave method is when picking up fluctuation signal,
The surface (generally using pressing mode) for needing for sensor to be fixed on measured body, is contact pickoff signals mode.In tunnel
It when lining, may create the problem that the size of (1) sensor pressing dynamics also has a certain impact to test result；(2)
It is affected by testee surface smoothness, when tested without uneven surface or when having laitance and other impurities, test error becomes
Greatly；(3) sensor and the self-vibration characteristic of pickup system have an impact to test signal；
So it is a kind of based on audio frequency non-contact, mobile engineering lossless detection method " impact echo audio method,
Impact Acoustic Echo method, IAE " come into being.The starting point of IAE method is exactly that sound pick-up is utilized to replace pick-up
Sensor picks up the vibration signal of measured body, and introduces the analysis method of IE method (Impact echo).It is characterized in that: (1)
Signal stabilization, reliable using the non-contact pick device of signal, frequency domain is wide, to avoid the picking up signal and altogether of impact elasticity wave method
Vibration problem；(2) analysis method of impact elasticity wave method (Impact echo) is used, to solve test depth is shallow, objectivity is poor etc.
Problem.
But existing tunnel-liner non-destructive testing technology, it depends on and is manually analyzed and judged, mainly deposit
In following problems:
No matter elastic wave, ultrasonic wave, electromagnetic wave etc. are to carry out a or two a mesh according to the testing requirements of test object
The detection of parameter is marked, detection parameters are more single, and analysis method is equally single, and the relevant parameter of magnanimity can not really obtain benefit
With；
Existing detection mode is to the more demanding of operator, because thickness of tunnel linings is uneven, and unknown, different technology
Personnel detect result that same object obtains using same equipment may be far from each other, and technical staff is needed to have skilled behaviour
Make skill and data abundant analysis experience.Meanwhile many judgements, there is no specific benchmark and threshold value, random and individual is poor
Greatly；
Meanwhile many detections need technical staff return to it is indoor by analysis software according to correlation step step by step and by
Data is analyzed, it is difficult to accomplish quick database preparation and Real-time Feedback test result.
Since the data of constituent parts, everyone are difficult to concentrate, verification result can not also summarize, the accuracy rate of test result and
Confidence interval is difficult to quantification, and objectivity is poor.
Summary of the invention
The purpose of the present invention is to provide a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning,
To solve the above problem in the prior art, realizes and improve detection accuracy, reduce the interference of personnel's subjectivity, it is ensured that testing result is objective
Accurate purpose.
The present invention is achieved through the following technical solutions:
A kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning, comprising the following steps:
(a) sensor is fixed in test object to and hit vibration, vibration signal is hit in acquisition；
(b) characteristic parameter is extracted from hitting in vibration signal for acquisition, the characteristic parameter includes reflection interval；Wherein to feature
Reflection interval in parameter carries out regression fit, obtains calibration value；
(c) original signal is indicated with obtained characteristic parameter, and according to the actual situation, this group of characteristic value is marked,
Its defect situation is recorded, in this, as a training set；
(d) step (a)~(d) is repeated in different test objects, increases training set quantity；
(e) model training is carried out using model training software: reading in all training sets first, then select corresponding classification
Device, and classifier parameters are set, it is modeled；
(f) by the model built up, the data of unknown testing result are parsed.
In the prior art, no matter elastic wave, ultrasonic wave, electromagnetic wave etc. according to the testing requirements of test object to carry out
The detection of a or two a target component, detection parameters are more single, and analysis method is equally single, and the relevant parameter of magnanimity can not be true
Just it is utilized；Existing detection mode is different because thickness of tunnel linings is uneven and unknown to the more demanding of operator
Technical staff detects result that same object obtains using same equipment may be far from each other, and technical staff is needed to have skillfully
Operation skill and data abundant analyze experience.Meanwhile much there is no specific benchmark and threshold values for judgement, it is random and a
People's difference is big；Meanwhile many detections need technical staff return to it is indoor by analysis software according to correlation step step by step and by
Data is analyzed, it is difficult to accomplish quick database preparation and Real-time Feedback test result.Due to constituent parts, everyone
Data be difficult to concentrate, verification result can not also summarize, and the accuracy rate and confidence interval of test result are difficult to quantification, objectivity
It is poor.In view of the above-mentioned problems, the present invention proposes a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning,
Sensor is fixed in test object to first and hit vibration, vibration signal is hit in acquisition；It is mentioned later from hitting in vibration signal for acquisition
Characteristic parameter is taken, the characteristic parameter includes reflection interval；Regression fit wherein is carried out to the reflection interval in characteristic parameter, is obtained
To calibration value.Since the variation of excavation and concrete material is crossed, owed in construction, so that reflection of the lining cutting back side to elastic wave
Time and design reflection interval are different.And designing reflection interval is major criterion parameter when linning defect detects, because
And error can be caused to determining defects.By analysis, either lining thickness is uneven or concrete material is uneven, produced by
The variation of elastic wave reflex signal all have roll-off characteristic.On the other hand, backside reflection signal caused by internal flaw prolongs
Then there is mutability afterwards.Therefore, regression fit is carried out to the reflection interval at the lining cutting back side in this method, in this, as baseline reflectance
Time, so as to avoid the influence of thickness, material variation.Regression fit herein can be according to any recurrence in the prior art
Approximating method is realized.Original signal is indicated with obtained characteristic parameter later, and according to the actual situation, to this group of characteristic value
It is marked, records its defect situation, in this, as a training set；In different test objects repeat step (a)~
(d), increase training set quantity.Model training is carried out using model training software later: being read in all training sets first, is then selected
Corresponding classifier is selected, and sets classifier parameters, is modeled；It is learned in this step using different training patterns
It practises, and then establishes the model that can judge lined internal defect, significantly improve detection accuracy.By the model built up, to not
Know that the data of testing result are parsed.This method due to cut down due to thickness, material variation generate adverse effect,
The reflection interval at the lining cutting back side is increased as determining defects parameter, therefore can preferably reflect the feature of defect, is solved
Existing detection method is affected by staff's subjective factor, the poor problem of detection accuracy, realizes raising detection
Precision reduces the interference of personnel's subjectivity, it is ensured that the objective and accurate effect of testing result.
Further, the characteristic parameter includes structure and boundary condition information, hits vibration signal message, reflection interval information
Ti, reflection interval information TiVariance rate RT between calibration valuei, reflection interval information TiVariance rate SRT between match valuei, with
And phase sensitivity index.
Further, the phase sensitivity index is Ti、RTi、Ti expBetween phase sensitivity index, wherein Ti expIt is bent according to fitting
The predicted time of line.
Further, the method for carrying out regression fit to reflection interval is m-1 regression fit, linear regression fitting, two
Secondary regression fit or multiple regression fitting.Other than the regression fit method listed in this programme, remaining time in the prior art
Approximating method is returned to be equally applicable in the application.
Further, when carrying out regression fit to reflection interval, discrete excessive data are removed.It can will be discrete excessive
Data remove after be fitted again.It is of course also possible to use being first fitted, then discrete excessive point is cast out, again
The mode being fitted.
Further, the discrete excessive data are the numerical value beyond 2 times of standard deviations of mean value ±.
Further, describedWherein TdFor the bottom reflection time theory value calculated according to design thickness；
When design thickness is unknown, RTiTake measured time parameter.
Further, describedWherein Ti expFor according to the predicted time of matched curve.Using m-1
When secondary regression fit mode,
Further, it is trained in step (e) by following any model: random forest, Bayesian network, neuron
Network.
Random forest is exactly a kind of algorithm by the thought of integrated study that more trees is integrated, its basic unit is to determine
Plan tree, and its essence belongs to a big branch of machine learning --- integrated study (Ensemble Learning) method.It is special
Point has: having fabulous accuracy rate；It can effectively operate on large data sets；It is capable of handling the input sample with high dimensional feature
This, and do not need dimensionality reduction；Importance of each feature in classification problem can be assessed；Default value problem can also be obtained
It obtains and obtains result very well；
Bayesian network is also known as belief network, is the extension of Bayes method, is current uncertain knowledge expression and reasoning neck
One of most effective theoretical model in domain.Bayesian network is suitable for expression and analysis uncertainty and probabilistic event, application
In the decision for conditionally relying on various control factor, can be done from incomplete, inaccurate or uncertain knowledge or information
Reasoning out.
Neuron net is a kind of mathematical model having used for reference the working principle of biological neural network and having been formed, it can both be used to
The task of supervision, such as classification, visual identity are done, unsupervised task is also used as.It is capable of handling complexity simultaneously
Nonlinear problem, its basic structure are neurons.
Further, the reflection interval information TiFundamental frequency cycles including lining cutting backside reflection time, reflection interval.
Compared with prior art, the present invention having the following advantages and benefits:
A kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning of the present invention, has cut down due to thickness
The adverse effect that degree, material variation generate, increases the reflection interval at the lining cutting back side as determining defects parameter, therefore can be compared with
The feature for reflecting defect well solves existing detection method and is affected by staff's subjective factor, detection accuracy
Poor problem realizes and improves detection accuracy, the interference of reduction personnel's subjectivity, it is ensured that the objective and accurate effect of testing result.
Detailed description of the invention
Attached drawing described herein is used to provide to further understand the embodiment of the present invention, constitutes one of the application
Point, do not constitute the restriction to the embodiment of the present invention.In the accompanying drawings:
Fig. 1 is the Bayesian network model schematic diagram in the specific embodiment of the invention；
Fig. 2 is the ARTOICAL NEURAL NETWORK MODEL schematic diagram in the specific embodiment of the invention；
Fig. 3 is the regression fit result schematic diagram in the specific embodiment of the invention to reflection interval.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, below with reference to embodiment and attached drawing, to this
Invention is described in further detail, and exemplary embodiment of the invention and its explanation for explaining only the invention, are not made
For limitation of the invention.
Embodiment 1:
A kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning, comprising the following steps:
(a) sensor is fixed in test object to and hit vibration, vibration signal is hit in acquisition；
(b) characteristic parameter is extracted from hitting in vibration signal for acquisition, the characteristic parameter includes reflection interval；Wherein to feature
Reflection interval in parameter carries out regression fit, obtains calibration value；
(c) original signal is indicated with obtained characteristic parameter, and according to the actual situation, this group of characteristic value is marked,
Its defect situation is recorded, in this, as a training set；
(d) step (a)~(d) is repeated in different test objects, increases training set quantity；
(e) model training is carried out using model training software: reading in all training sets first, then select corresponding classification
Device, and classifier parameters are set, it is modeled；
(f) by the model built up, the data of unknown testing result are parsed.
Wherein, there is N number of measuring point, the distance of measuring point i is xi, corresponding reflection interval is yi；
If had using m-1 regression fit:
It such as returns, then has using straight line is (primary):Wherein, wherein For mean value.
Such as using quadratic regression is used, then have:
Wherein, y0For the y value of first measuring point.
In actual detection engineering, the design thickness of the lining cutting (especially two linings) of detection interval is definite value.But due to
There may be defects etc. for lined internal, so that there is variation by a relatively large margin in reflection interval.Therefore, it in above-mentioned fitting, can incite somebody to action
The data of discrete excessive (as exceeded 2 times of standard deviations of mean value ±) are fitted again after removing.It is of course also possible to use first carrying out
Fitting, then discrete excessive point is cast out, be fitted again.
Embodiment 2:
A kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning, on the basis of embodiment 1, institute
Characteristic parameter is stated to include structure and boundary condition information, hit vibration signal message, reflection interval information Ti, reflection interval information TiWith
Variance rate RT between calibration valuei, reflection interval information TiVariance rate SRT between match valueiAnd phase sensitivity index.The phase sensitivity
Index is Ti、RTi、Ti expBetween phase sensitivity index, wherein Ti expFor according to the predicted time of matched curve.To reflection interval into
The method of row regression fit is m-1 regression fit, linear regression fitting, Quadratic Regression Fitting or multiple regression fitting.To anti-
When penetrating time progress regression fit, discrete excessive data are removed.The discrete excessive data are to mark beyond 2 times of mean value ±
The numerical value of quasi- deviation.It is describedWherein TdFor the bottom reflection time theory value calculated according to design thickness；When setting
When meter thickness is unknown, RTiTake measured time parameter.It is describedWherein Ti expFor according to the prediction of matched curve
Time.It is trained in step (e) by following any model: random forest, Bayesian network, neuroid.It is described anti-
Penetrate temporal information TiFundamental frequency cycles including lining cutting backside reflection time, reflection interval.
Fig. 3 be the present embodiment to the regression fit of reflection interval as a result, RefC is the lining cutting backside reflection time in figure；FFT0
For FFT (Fast Fourier Transform (FFT)) frequency spectrum fundamental frequency cycles；MEM0 is MEM (maximum entropy method (MEM)) frequency spectrum fundamental frequency cycles.
The present embodiment according to the shape of test object, the situation of working face, selects optimal test method first, it is known that
Defect is as shown in table 1:
Lining thickness defect verify data list known to table 1
For upper table, inventor uses 2 layers of Bayesian network and neuron net as shown in Figure 2 as shown in Figure 1
Network classifier, and it is added to assembled classifier, verification mode is verified using right-angled intersection and full training set verifies two ways,
Obtain result as shown in Table 2:
The precision of the various identification models of table 2 compares
From Table 2, it can be seen that by the obtained modeling result of this method, the accuracy rate energy in terms of detecting linning defect
Enough reach ninety percent or more, accuracy rate is high, and not by the interference of staff.It will be evident that all the precision after training is higher than friendship
Training precision is pitched, therefore specific to being exactly that training set quantity is more, and model accuracy is higher in the application, to unknown testing result
The parsing result of data is more accurate.
Above-described specific embodiment has carried out further the purpose of the present invention, technical scheme and beneficial effects
It is described in detail, it should be understood that being not intended to limit the present invention the foregoing is merely a specific embodiment of the invention
Protection scope, all within the spirits and principles of the present invention, any modification, equivalent substitution, improvement and etc. done should all include
Within protection scope of the present invention.
Claims (10)
1. a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning, which is characterized in that including following step
It is rapid:
(a) sensor is fixed in test object to and hit vibration, vibration signal is hit in acquisition；
(b) characteristic parameter is extracted from hitting in vibration signal for acquisition, the characteristic parameter includes reflection interval；Wherein to characteristic parameter
Interior reflection interval carries out regression fit, obtains calibration value；
(c) original signal is indicated with obtained characteristic parameter, and according to the actual situation, this group of characteristic value is marked, recorded
Its defect situation, in this, as a training set；
(d) step (a)~(d) is repeated in different test objects, increases training set quantity；
(e) model training is carried out using model training software: reading in all training sets first, then select corresponding classifier,
And classifier parameters are set, it is modeled；
(f) by the model built up, the data of unknown testing result are parsed.
2. a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning according to claim 1,
It is characterized in that, the characteristic parameter includes structure and boundary condition information, hits vibration signal message, reflection interval information Ti, reflection
Temporal information TiVariance rate RT between calibration valuei, reflection interval information TiVariance rate SRT between match valueiAnd phase sensitivity
Index.
3. a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning according to claim 2,
It is characterized in that, the phase sensitivity index is Ti、RTi、Ti expBetween phase sensitivity index, wherein Ti expFor according to the prediction of matched curve
Time.
4. a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning according to claim 1,
It is characterized in that, it is quasi- for m-1 regression fit, linear regression fitting, quadratic regression to carry out the method for regression fit to reflection interval
It closes or multiple regression is fitted.
5. a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning according to claim 1,
It is characterized in that, when carrying out regression fit to reflection interval, discrete excessive data is removed.
6. a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning according to claim 5,
It is characterized in that, the discrete excessive data are the numerical value beyond 2 times of standard deviations of mean value ±.
7. a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning according to claim 2,
It is characterized in that, it is describedWherein TdFor the bottom reflection time theory value calculated according to design thickness；When design is thick
When spending unknown, RTiTake measured time parameter.
8. a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning according to claim 2,
It is characterized in that, it is describedWherein Ti expFor according to the predicted time of matched curve.
9. a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning according to claim 1,
It is characterized in that, is trained in step (e) by following any model: random forest, Bayesian network, neuroid.
10. a kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning according to claim 2,
It is characterized in that, the reflection interval information TiFundamental frequency cycles including lining cutting backside reflection time, reflection interval.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910317985.3A CN110082429B (en) | 2019-04-19 | 2019-04-19 | Tunnel lining nondestructive testing auxiliary judgment method combining machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910317985.3A CN110082429B (en) | 2019-04-19 | 2019-04-19 | Tunnel lining nondestructive testing auxiliary judgment method combining machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110082429A true CN110082429A (en) | 2019-08-02 |
| CN110082429B CN110082429B (en) | 2022-02-22 |
Family
ID=67415613
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910317985.3A Active CN110082429B (en) | 2019-04-19 | 2019-04-19 | Tunnel lining nondestructive testing auxiliary judgment method combining machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110082429B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110569917A (en) * | 2019-09-16 | 2019-12-13 | 四川升拓检测技术股份有限公司 | sleeve grouting compactness discrimination method based on deep learning image recognition |
| CN110688698A (en) * | 2019-09-25 | 2020-01-14 | 东北大学 | Intelligent surrounding rock large deformation assessment method based on random forest algorithm |
| CN110736980A (en) * | 2019-11-04 | 2020-01-31 | 广东博智林机器人有限公司 | method, device and system for monitoring climbing rack state information |
| CN111487315A (en) * | 2020-04-17 | 2020-08-04 | 四川陆通检测科技有限公司 | Audio frequency nondestructive testing method for tunnel lining thickness and void |
| CN111812149A (en) * | 2020-07-20 | 2020-10-23 | 南京工业大学 | Adiabatic acceleration calorimetry method based on machine learning |
| CN113705289A (en) * | 2021-01-23 | 2021-11-26 | 深圳市玄羽科技有限公司 | Method for improving nondestructive testing precision based on machine learning |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108491931A (en) * | 2018-03-29 | 2018-09-04 | 四川升拓检测技术股份有限公司 | A method of non-destructive testing precision is improved based on machine learning |
-
2019
- 2019-04-19 CN CN201910317985.3A patent/CN110082429B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108491931A (en) * | 2018-03-29 | 2018-09-04 | 四川升拓检测技术股份有限公司 | A method of non-destructive testing precision is improved based on machine learning |
Non-Patent Citations (2)
| Title |
|---|
| 景喜林等: "探地雷达在铁路隧道衬砌质量检测中的应用", 《工程地球物理学报》 * |
| 王梦恕: "《隧道与地下工程技术及其发展》", 31 December 2004, 北京交通大学出版社 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110569917A (en) * | 2019-09-16 | 2019-12-13 | 四川升拓检测技术股份有限公司 | sleeve grouting compactness discrimination method based on deep learning image recognition |
| CN110688698A (en) * | 2019-09-25 | 2020-01-14 | 东北大学 | Intelligent surrounding rock large deformation assessment method based on random forest algorithm |
| CN110688698B (en) * | 2019-09-25 | 2024-01-30 | 东北大学 | Surrounding rock large deformation intelligent evaluation method based on random forest algorithm |
| CN110736980A (en) * | 2019-11-04 | 2020-01-31 | 广东博智林机器人有限公司 | method, device and system for monitoring climbing rack state information |
| CN111487315A (en) * | 2020-04-17 | 2020-08-04 | 四川陆通检测科技有限公司 | Audio frequency nondestructive testing method for tunnel lining thickness and void |
| CN111812149A (en) * | 2020-07-20 | 2020-10-23 | 南京工业大学 | Adiabatic acceleration calorimetry method based on machine learning |
| CN111812149B (en) * | 2020-07-20 | 2021-10-15 | 南京工业大学 | Adiabatic acceleration calorimetry method based on machine learning |
| CN113705289A (en) * | 2021-01-23 | 2021-11-26 | 深圳市玄羽科技有限公司 | Method for improving nondestructive testing precision based on machine learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110082429B (en) | 2022-02-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110082429A (en) | A kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning | |
| Márquez et al. | A review of non-destructive testing on wind turbines blades | |
| CN106990018B (en) | A kind of prestressed concrete beam Grouted density intelligent identification Method | |
| Liu et al. | Overall grouting compactness detection of bridge prestressed bellows based on RF feature selection and the GA-SVM model | |
| CN109190272B (en) | Concrete structure defect detection method based on elastic waves and machine learning | |
| CN109374631A (en) | A kind of tunnel state evaluating method | |
| CN115390033A (en) | Method, system and device for detecting cavity and water loss under road surface based on ground penetrating radar | |
| CN116401571A (en) | Concrete filled steel tube abnormal shape void identification method based on knocking sound wave and MiniRoccket | |
| He et al. | A fast acoustic emission beamforming localization method based on Hilbert curve | |
| Yu et al. | Automated health condition diagnosis of in situ wood utility poles using an intelligent non-destructive evaluation (NDE) framework | |
| Martino et al. | Determining ground penetrating radar amplitude thresholds for the corrosion state of reinforced concrete bridge decks | |
| CN113468804B (en) | Underground pipeline identification method based on matrix bundles and deep neural network | |
| CN115616674A (en) | Ground penetrating radar training data set simulation amplification and road nondestructive testing method and system | |
| Zhang et al. | Automatic corrosive environment detection of RC bridge decks from ground-penetrating radar data based on deep learning | |
| Wickramanayake et al. | Deep learning for estimating low-range concrete sub-surface boundary depths using ground penetrating radar signals | |
| Zhang et al. | THz Wave Detection of Gap Defects Based on a Convolutional Neural Network Improved by a Residual Shrinkage Network | |
| JP2020051851A (en) | Abnormal state detection method within reinforcing-bar concrete structure | |
| CN110231401B (en) | Inter-measuring point information correlation method and civil structure nondestructive testing method | |
| Küttenbaum et al. | Ways to unlock the potential of non-destructive concrete testing for the reliability assessment of our built environment | |
| Guo et al. | Research on tunnel lining image target recognition method based on YOLOv3 | |
| Wong et al. | Harmonising Cross-System GPR Wave Amplitude for Concrete Diagnosis with Machine Learning | |
| Barbosh et al. | Damage identification in concrete structures using a hybrid time–frequency decomposition of acoustic emission responses | |
| KR102611457B1 (en) | AI-based defect detection system inside concrete members | |
| CN116910489B (en) | Wall seepage prevention detection method based on artificial intelligence and related device | |
| Mizutani et al. | Signal processing for fast RC bridge slab damage detection by using UHF-band radar |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |